20:17
[user] I think it exists already, it's just called mold
[that mac] but it accepts another matrix iirc
[RubenVerg] oh right forgot about mold
[RubenVerg] doesn't do a reshape though
[RubenVerg] it does reshape-to-shape-of
[that mac] yeah that's what I meant
[that mac] would reshape be better suited as a digraph?
@pacman256 [user] If we add an element to do this, then reshape would essentially be a two-byter
[RubenVerg] I think it makes more sense to add an element for reshape which makes the matrix essentially a two-byter
[RubenVerg] or maybe three-byter, zero wrap reshape
[that mac] hmm but zero matrix feels like an important element too
[user] That works too, idk which one is more commonly used
[RubenVerg] zero matrix to me sounds like a specific application of reshape
[RubenVerg] also mold is a weird primitive, it's like a less powerful combo of shape + reshape
[that mac] maybe nuke mold?
[that mac] and replace it with shape-of and reshape or smth?
[RubenVerg] well I guess mold might be more general in that it probably works with non-rectangular arrays
[RubenVerg] like a mold on [[1, 2, 3], [4, 5]] I assume works
[RubenVerg] whereas there is no way to spell that in the language of shape/reshape
@pacman256 [RubenVerg] I think ʀ/ᵒ0 might work, though only with matrices (not with higher-rank arrays)
[RubenVerg] ʀ vectorizing on lists instead of doing APL-like multidimensional index generator is so cursed
[RubenVerg] maybe a multidimensional range element could be added?
[RubenVerg] or maybe even replace current range-creation elements, I doubt any code depends on range-creation vectorization
